Update app.py

app.py

@@ -22,7 +22,7 @@ def predict_clothing(images):
     messages = [{"role": "user", "content":
                  [
                      {"type": "image"},
-                     {"type": "text", "text": "Define this clothing in 1-3 words. Your response should be only the definition."}
+                     {"type": "text", "text": "Define this clothing in 1-3 words, always include the colour of the clothing too. Your response should be only the definition."}
                  ]}
                ]
     input_text = processor.apply_chat_template(messages, add_generation_prompt=True)
@@ -57,10 +57,10 @@ def generate_image(category, img1, img2, img3, height, width, img_guidance_scale
     wears = predict_clothing(input_images[1:])
     if len(wears)==1:
         dress = wears[0]
-        text = f"""A {category} posing straight in a {dress}. The {category} is in <img><|image_1|></img>. The {dress} is in <img><|image_2|></img>."""
+        text = f"""The {category} in <img><|image_1|></img> in a {dress}. The {dress} is in <img><|image_2|></img>."""
     elif len(wears)==2:
         topwear, bottomwear = wears[0], wears[1]
-        text = f"""A {category} posing straight in a {topwear} and a {bottomwear}. The {category} is in <img><|image_1|></img>.
+        text = f"""The {category} in <img><|image_1|></img> in a {topwear} and a {bottomwear}.
                     The {topwear} is in <img><|image_2|></img>. The {bottomwear} is in <img><|image_3|></img>."""
     else:
         input_images = None